| Network
of Excellence |
CARiMan |
The
Internet Portal for Computer Aided Risk Management |
Prof.
Alfredo Goņi
University
of the Basque Country
Facultad de Informática
Paseo Manuel de Lardizábal 1
(20018)
Donostia-San Sebastián
Spain
T: (34) 943015066
Fax: (34) 943219306
alfredo@si.ehu.es
The
University of the Basque Country (Spain) is, the only Public University in the
Basque Country Autonomous Community, and the one with the widest range
educational offer, which almost
reaches
one hundred qualifications. In its thirty faculties and university schools,
placed through the campuses
of Araba, Bizkaia and Gipuzkoa, work, study and investigate more than 60,000
students, 3,500 lectures and a thousand professional staff. The University´s
anagram, which was designed by the artist Eduardo Chillida, is joined to the
motto "Eman ta zabal zazu", "Bear fruit and make it known".
This is the leading principle of all the university activity.
The
Interoperable Database Group at the Computer Languages and Systems Department of
the University of the Basque Country (http://siul02.si.ehu.es/), that
participates in this proposal, has been involved in several research projects
that fit in the following research issues:
ˇ
Data
Management Issues in Mobile
Computing
ˇ
Heterogeneous,
Distributed Information Systems
ˇ
Query
Processing on Global Information Systems
ˇ
Terminological
Systems / Description Logics
At
this moment we
try to apply to the area of telemedicine, our background in the fields of: 1)
data management (in heteregeneous, distributed, federated databases and global
information systems); 2) reasoning systems like those based on KL-ONE, also
known as Terminological Systems or systems based on Description Logics; and, 3)
new data services for mobile computing. In particular, our main research field
at the moment is ubiquitous monitoring of people, more concretely in two
subareas of monitoring: assistance of elderly people and ECG monitoring of
patients, focusing mainly in the
data management issues related with these subareas of the telemedicine.
Description of the work performed by our group at the University of the Basque Country:
We are working in two lines related to telemedicine:
1) Classifying ECGs in an
On-Line Monitoring System
2) Tele assistance of elderly people
1) Classifying ECGs in an
On-Line Monitoring System
PATIENTS with heart rhythm irregularities which are not
detected on a normal electrocardiogram (ECG) require
some type of ambulatory detection using a holter. The use of a
holter consists in placing electrodes (leads) on the patients'
chests; these leads are attached to the holter [1]. After the
patient is sent home and goes back to normal life, a tape
records a continuous ECG for 24 or 48 hours. One or two days
later, the holter is removed and the tape is analyzed. A
physician will see each of the patients' heart beats and if
abnormal beats or rhythms occur during that period, they are
identified by the physician. Although this solution presents the
advantage that patients can continue living a normal life in
their houses, it also presents a serious drawback: if the patient
suffers from a serious rhythm irregularity, the holter only
records it, i.e. it does not react to it.
In order to overcome the previous restriction, some new
proposals have appeared. We classify these proposals into two
major groups: commercial systems and research projects.
In the first group we include those commercial systems that
use a mobile telephone unit or a PDA (Personal Digital
Assistant) to capture the ECG signal. Three or four metalelectrodes, which record the heart event, are situated on the
back of a standard cellular phone. The data are transmitted to the cardiac monitor center situated at a hospital (e.g.
Vitaphone [2]). Another alternative is to use PDAs.
Companies like Ventracor [3] from Australia or Cardio Control [4] from the Netherlands have developed systems
capable of recording and storing ECGs directly in the PDA.
Additional features like GSM/GPRS transmission to an
analyzing unit are also being developed.
In the second group there are several research projects like
@Home [5], TeleMediCare [6] or PhMon [7], whose aims are
to build platforms for real time remote monitoring. These
systems include wireless bio-sensors that measure vital
parameters such as heart rate, blood pressure, insulin level, etc. The health monitoring system, carried by the patients, controls
these sensors. The patient data recorded are sent to the
hospital, where they are analyzed.
All the previous systems continuously send ECGs to the
hospital through a wireless communication network. The
monitoring of the whole signal takes place at the hospital. In
spite of the advantages these kinds of systems provide in
relation to holters, they still present main problems related to
the fact that the analysis is not performed in the place where
the signal is acquired. Therefore, there is a loss of efficiency in
the use of the wireless network because normal ECGs are also
sent (and wireless communications imply a high cost); and, in
the case of the wireless network is not available at some
moment, there might be a loss of ECG signal with the
corresponding risk of not detecting some anomalies. In order
to overcome these restrictions, we have built the MOLEC
system that analyzes the signals locally. In that case, although
the wireless communication is unavailable the signal is still
analyzed locally at the PDA, very close to the place where the
signal is acquired. Moreover, our system sends alarms to the
hospital when high-risk arrhythmias are found, so the cost of
the communication is less.
In the literature, there are some works which aim is to
classify beats and rhythms. For example: CARDINET,
CALICOT and Open Source. CARDINET [8] is a system of
on- line detection, classification and identification of the most
important wave of the beat: the QRS complex, in order to
classify the different types of heart beats. It detects the different types of beats with an error of 13% and a sensibility
of 83%. MOLEC only reaches an error of 3.872% classifying
beats and it also classifies rhythms that CARDINET does not.
CALICOT [10] (Cardiac Arrhythmias Learning for Intelligent
Classification of On-Line Tracks) analyzes the ECG signals in
order to identify arrhythmias by detecting representative
patterns. However, these patterns have not been confronted to
a real ECG database such as the well-known MIT-BIH [11]
MOLEC manages data of MIT-BIH. Open Source ECG
Analysis Software [9] analyzes ECG signals in real time in
order to detect Premature Ventricular Contractions (V), and
Bigeminy Arrhythmia (B). However, it only classifies one type
of beat and one type of rhythm, while MOLEC works with
fourteen types of beats and thirteen types of rhythms.
In summary in this paper, we present MOLEC, a continuous monitoring PDA-based system that records user ECG signals,
analyzes them into the PDA in order to find arrhythmias; and,
in case those high-risk arrhythmias are found, generates alarms
and sends them to cardiologists, so that they can determine
what to do with the user. Notice that in this approach, there is
no a continuous transmission of ECGs to the hospital.
Moreover, instead of using some of the existing proposals in
order to classify beat and arrhythmias, we decided to define
new ones for the following reasons. Firstly, many of them
classify only one specific type of beats/rhythms (e.g. Open
Source). Secondly, their classifying precision does not reach
100% (e.g. CARDINET). Lastly, they are not oriented to a
PDA framework (e.g. all the previously mentioned).
In order to obtain a most accurate beat and rhythm
classifier, we compared several tools and methods, in machine
learning area. Among those methods we can mention:
Bayesian networks or decision trees [12], knowledge-based
approaches that use expert rules [13-14], or fuzzy logic [15].
In the rest of the paper we present an overview of the
MOLEC system. Afterwards, we describe the set of
experiments performed to find the most accurate classifier.
Then, the obtained results using the chosen classifier are
shown for beat and rhythm classification. Lastly, we finish
with our conclusions.
2) Tele assistance of elderly people
Population
surveys agree in the following statement:
First
worlds countries are getting older. It is a reality
that
the number of elderly people is growing and so, the
number
of these people that live alone is increasing. Although
many
of them can manage themselves, it is also true that they
feel
a kind of defenceless situation and many of them sign a
contract
with companies that offer a tele assistance service.
Most
of tele assistance services offered nowadays are based
on
hardware equipment located at the elderly person home
which
is connected to the wired public phone network. There
is
also a kind of medallion that the person wears on the wrist or
around
the neck and a system (software and hardware) located
at
the company office that provides the tele assistance service.
When
the person feels bad (physically or emotionally) she
presses
the medallion and a conversation link is established
with
a person of the tele assistance company. This operator
decides
the measures to take.
These
tele assistance services, although they accomplish an
interesting
and necessary function, present the following main
constraints:
1)
They are passive, i.e., they mainly react only when the
user
requires it.
2)
Their coverage is limited, normally constrained to the
persons
home and to a radius of some meters around
the
home.
3)
They do not monitor automatically vital signs that, in
some
situations, may have repercussion on the persons
life.
Having
as a goal to overcome the previous constraints, we
are
developing the system AINGERU, that is our proposal for
a
new way of tele assistance for elderly people. AINGERU
takes
benefit from the new advances in the areas of networking
(wireless
communications), mobile computing (Personal Digital
Assistants)
and artificial intelligence (semantic web and
agent technologies) to accomplish its goal.
Hence,
apart from supporting the functionalities provided
by
present tele assistance services, AINGERU also:
_
offers
an active assistance by using agents that behave
in
the face of anomalous situations without a direct
intervention
of the user.
_
offers
an anywhere and anytime assistance by using
wireless
communications and PDAs.
_
and
allows to monitor vital signs by using sensors that
capture
the values of those signs and feed a decision
support
system that analyzes them and generates an alarm
when
necessary.
Although
a detailed description of the global AINGERU
system
goes beyond the scope of this paper, what we are going
to
show are the main features of the use of three technologies:
agents,
semantic web and web services. The combined use of
this
three technologies constitutes the core that differentiate
AINGERU
from other related works.
1)
Agents: We have
developed several agents that run in
different
elements that are part of the AINGERU system (for
example,
in the PDA that the person wears, in the Control
Center
that is in charge of monitoring people, etc.). The main
advantages
that the use of the agent technology provides to
AINGERU
are:
_
agents
help to manage the distributed and heterogeneous
framework
of AINGERU (a big number of PDAs, several
Control
Centers, several Care Centers, several Health
Centers,
etc.).
_
agents
can accomplish tasks (for example, to monitor
vital
signs) autonomously and they can interact with other
agents
(for example, the Condition Checker Agent contacts
Emergency
Agent to dispatch an alarm) to perform
complex
tasks (in an alarm management, there are more
than
five cooperating agents involved).
_
they
can learn from the user behaviour and so, they can
be
self adaptative.
2)
Semantic web: We
have developed two logic based
ontologies,
MedOnt and OperOnt, that allow reasoning on
them.
MedOnt describes the different situations in which a
medical
alarm has to be activated. Hence, in this ontology, not
only
the different symptoms that a user can have are described,
with
respect to vital signs that several sensors can monitor,
but
also the usual illnesses that elderly people suffer from.
Moreover,
this ontology can be customized for every user.
The
purpose of the second ontology, OperOnt, is to describe
the
operational model of AINGERU. Notice that the
development
of this ontology is also a novelty that AINGERU
brings
with respect to other related systems. Terms in this
ontology
describe types of agents, types of messages for the
communication
among agents, types of web services, etc.
Those
terms are defined independently from any agent system
implementation,
so that it helps interoperability among agents.
At
the same time, the OperOnt ontology describes contextual
information
that several agents are able to share. This ontology
is
easily extended as the functionality of AINGERU increases.
3)
Web services: Web
services are used to integrate non
agent
oriented applications into AINGERU. Using those web
services,
outside applications can feed data into AINGERU or
retrieve
data from it. Moreover, there has been also developed
a
method that provides HTML access, allowing relatives and
physicians
related to a monitored person, using a browser, to
consult
data about user appointments, medicines taken, etc.
All
the web services are described in the OperOnt ontology
in
order to represent them at the semantic level.
The
members of the Interoperable Database Group at the University of the Basque
Country are:
Alfredo
Goņi, associate professor at the Computer Languages and Systems Department of
the University of the Basque Country. He has a BS degree and a PHD at Computer
Science.
Arantza
Illarramendi, professor at the Computer Languages and Systems Department of the
University of the Basque Country. She has a BS degree and a PHD at Computer
Science.
Jesús
Bermúdez, associate professor at the Computer Languages and Systems Department
of the University of the Basque Country. He has a BS degree and a PHD at
Computer Science.
Jimena
Rodríguez, research assistant at the Computer Languages and Systems Department
of the University of the Basque Country. She has a BS degree at Computer Science
and currently is a PhD candidate.
Alberto
Tablado, research assistant at the Computer Languages and Systems Department of
the University of the Basque Country. He has a BS degree at Computer Science and
currently is a PhD candidate.
SELECTED
PUBLICATIONS (Complete publications in http://siul02.si.ehu.es/)
1.
A.
Tablado, A. Illarramendi, J. Bermúdez and A. Goņi. Intelligent Monitoring of
Elderly People. The 4th Annual IEEE EMBS Special Topic Conference on Information
Technology Applications in Biomedicine (ITAB 2003), Birmingham (UK), April 2003.
2.
J.
Rodríguez, A. Goņi and A. Illarramendi. On-Line Management of User Heart Data
to Detect Anomalous Behavior. Sent to 29th VLDB Conference (Very Large
Databases), Berlin 2003
3.
E.
Mena, J.A. Royo, A. Illarramendi and A. Goņi, An Agent-based Approach for
Helping Users of Hand-Held Devices to Browse Software Catalogs,
Cooperative Information Agents VI, 6th International Workshop CIA 2002, Madrid
(Spain), Lecture Notes on Artificial Intelligence (LNAI), ISBN 3-540-44173-5,
pp. 51-65, September 2002.
4.
Y.
Villate, A. Illarramendi and E. Pitoura, Keep Your Data Safe and Available While
Roaming, International Journal of Mobile Networks and Application (MONET),
Special Issue on Pervasive Computing, 7(4):315-328, August 2002.
5.
J.
Rodriguez, Y. Villate, A. Goņi, A. Illarramendi and E. Mena, Data Services for
Wireless Devices: from laptops to PDAs and from GSM to GPRS, Wireless
Information Systems. Proceedings of the First International Workshop on Wireless
Information Systems (WIS 2002), Ciudad Real (Spain), ISBN. 972-98816-0-X, Qusay
H. Mahmoud (ed.), ICEI Press, pp. 70-81, April 2002.
6.
A.
Goņi, A. Illarramendi, E. Mena, Y. Villate and J. Rodriguez, ANTARCTICA: A
Multiagent System for Internet Data Services in a Wireless Computing Framework.
Developing an Infrastructure for Mobile and Wireless Systems. Lecture Notes in
Computer Science (LNCS 2538), ISBN 3-540-00289-8, pp. 119-135, 2002.
7.
E.
Mena and A. Illarramendi, Ontology-Based Query Processing for Global Information
Systems, Kluwer Academic Publishers, ISBN 0-7923-7375-8, pp. 215, 2001. June
2001.
8.
A.
Goņi and A. Illarramendi, Mobile Computing: Data Management Issues, chapter of
the book "Advanced Database Technology and Design", M. Piattini and O.
Díaz (ed.), Artech-House 2000.
9.
E.
Mena, V. Kashyap, A. Illarramendi and A. Sheth, Imprecise Answers on Highly Open
and Distributed Environments: An Approach based on Information Loss for
Multi-Ontology Based Query Processing, International Journal of Cooperative
Information Systems (IJCIS), special issue Intelligent Integration of
Information, 9(4):403-425, December 2000.
10.
E.
Mena, A. Illarramendi, V. Kashyap and A. Sheth, OBSERVER: An Approach for Query
Processing in Global Information Systems based on Interoperation across
Pre-existing Ontologies, International journal on Distributed And Parallel
Databases (DAPD), ISSN 0926-8782, 8(2):223-272, April 2000.
11.
J.M.
Blanco, A. Goņi and A. Illarramendi, Mapping among Knowledge Bases and
Databases: Precise definition of its syntax and semantics, Information Systems,
ISSN 0306-4379, 24(4):275-301, Elsevier Science Ltd., 1999.
12.
A.
Goņi, A. Illarramendi, E. Mena and J.M. Blanco, An Optimal Cache for a
Federated Database System, Journal of Intelligent Information Systems (JIIS),
ISSN 0925-9902, 9(2):125-156, Kluwer Academic Publishers, September/October
1997.
13.
E.
Bertino and A. Illarramendi, The Integration of Heterogeneous Data Management
Systems: Approaches based on the Object-Oriented paradigm, chapter of the book
"Object Oriented Multidatabase Systems", O. Bukhres and E. Elmagarmid
(ed.), Prentice-Hall, 1995.
14.
J.M.
Blanco, A. Illarramendi and A. Goņi, Building a Federated Database System: an
approach using a Knowledge Based System, International Journal on Intelligent
and Cooperative Information Systems, 3(4):415-455, December 1994.
15.
A.
Illarramendi, J.M. Blanco and A. Goņi, Making the Knowledge Base Systems More
Efficient: A Method to Detect Inconsistent Queries, IEEE Transactions on
Knowledge and Data Engineering, 6(4):634-639, August 1994.
| Network
of Excellence |
CARiMan |
The
Internet Portal for Computer Aided Risk Management |